STAC Summit, 15 Nov 2018, London

STAC Summits

STAC Summits bring together industry leaders in architecture, app dev, infrastructure
engineering, and operational intelligence to discuss important technical challenges in
the finance industry. Come to hear leading ideas and exchange views with your peers.

America Square Conference Center
17 Crosswall, London EC3N 2LB


Click on the session titles to view the slides.


Big Compute Big Compute

Fast Compute Fast Compute

Big Data Big Data

Fast Data Fast Data


Women in TechnologyFast Data   

It doesn’t take a statistician to figure out that women occupy relatively few senior positions in technology today. This seems even truer for technology in the finance industry. Yet some women are clearly succeeding. In this panel, we’ll ask women who have enjoyed success in technology-intensive roles to discuss the keys to their success, how they view the current situation for women in tech, and how financial firms or tech vendors can get a leg up in the war for talent by facilitating female careers.

Goodbye, Data Lake: Why continuous analytics yield higher ROIFast DataBig Data   Fast Compute   

Faced with the need to handle increasing volumes of data, alternative datasets ("alt data"), and AI, many enterprises are working to design or redesign their big data architectures. A traditional approach is to store everything in a data lake, process it via ETL, and analyze it in batch or interactive modes. However, in Ori's view, this decade-old approach fails to generate sufficient ROI. In this talk, he will argue for a different approach in which information is ingested, enriched, and analyzed in context as it arrives, including via machine learning or deep learning, then immediately made available to users or to drive automated actions. He will also argue that it's possible to take full advantage of modern hardware, micro-services and serverless functions to achieve much higher performance while still benefiting from CI/CD, auto-scaling, and fast software rollouts. In Ori's view, the resulting "continuous analytics" solutions yield faster answers for the business while remaining simpler and less expensive for IT.

Update on Big Workloads and CloudBig Data   Big Compute   

Peter will present the latest benchmark results for big data workloads like tick analytics and big compute workloads like derivatives valuation. He will also provide an update on the STAC Cloud Special Interest Group SIG, a group of financial firms and vendors focused on improving assessments and dialog relating to public, private, and hybrid cloud solutions.

Innovation RoundupBig Data   Big Compute   
  "Levyx: Trends in Low Latency Analytics and Computational Storage"
    Matt Meinel, SVP Solutions Architecture, Levyx
  "Streaming Analytics Tools on FabricXpress: Changing Performance in the Financial Sector"
    Hollis Beall, Director of Performance Engineering, Axellio
  "Bending the Rules of Reality for Faster Data Analytics"
    Andy Fleisch, Regional Sales Manager, Weka.IO


Panel: Staying ahead of data analytics challengesBig Data   Big Compute   

The challenges inherent in data analytics to support trading and investment, as well as the potential of new technologies to help, have been big topic of research and discussion at STAC for about a decade. Even after all that time, it's still a hot topic. Why is that? What are the most recent challenges, and how can firms solve them? Our panel will bring together several innovators with unique angles on the topic. To kick off, the vendors on the panel will each provide a short presentation:

  "Run your code in the database: User Defined Functions in Python"
   Jean-Claude Tagger
  "Bridging the Gap: Bigstream Hyper-Acceleration for Data Analytics"
   Roop Ganguly
  "Tackling the challenges of rapidly growing data stores"
   Ayelet Heyman


Benchmarking the value drivers of ML solutionsBig Data   Big Compute   

It's a non-trivial challenge to benchmark machine-learning techniques and technologies in a way that captures the key elements that are important to decision makers in a business. Investment managers and trading desks care about many things beyond raw performance, such as model quality, time to market, and cost. In this talk, Michel will use early test results from a common text-processing use case in finance to illustrate a general framework for ML benchmarks that can be applied by financial firms and technology vendors (oh, and by benchmarking firms).

Innovation RoundupBig Data   Big Compute   
  "Lenovo AI/ML Innovations"
    Dave Weber, Wall Street CTO & Director, Lenovo
  "Delivering Real Time Risk on FPGAs"
    Oskar Mencer, CEO, Maxeler Technologies


Is FPGA acceleration of financial analytics viable?Fast DataBig Data   Fast Compute   Big Compute   

FPGA technology has traditionally required custom hardware design and development in RTL. But according to panelists at the STAC Summits last spring, hardware and software is now available to ease creation of FPGA-accelerated algorithms and to deploy them quickly in the datacenter. The question is: are these new products up to the challenging demands of financial analytics? In this talk, Intel will discuss what kinds of workload make good targets for FPGA and present recent case studies where FPGA acceleration has been used for financial applications. While the main focus of these cases is the acceleration achieved, Intel will also discuss other key attributes such as scaling and ease of use.

“Different” doesn’t mean “Difficult”: FPGA programming demystifiedFast DataBig Data   Fast Compute   Big Compute   

FPGAs are emerging in more and more places as a way to accelerate a wide range of workloads, including financial analytics. Unlike traditional processing devices such as CPUs and GPUs, FPGAs are not instruction set machines with fixed architectures. Instead, FPGAs enable developers to create custom architectures optimized for specific workloads. To take advantage of this flexibility, developers must approach FPGA programming differently. But as Sergei puts it, “different” doesn’t necessarily mean “difficult”. Sergei will argue that FPGA programming is well within the reach of most software programmers, thanks to familiar languages such as C, C++, and OpenCL. The critical thing, according to Sergei, is understanding instruction-level, data-level, and task-level parallelism, along with the dataflow paradigm. In this talk, Sergei will detail how new software-like development flows are making FPGA acceleration accessible to a much wider audience.

STAC Update: Fast DataFast Data

Peter will discuss the latest research and Council activities related to low-latency/high-throughput realtime workloads, including time synchronization.

Innovation RoundupFast Data   
  "Introducing NovaSparks Next Generation Platform"
    Yves Charles, VP Business Development, NovaSparks
  "Next-gen Market Access: ULL tick-to-trade made easier with Enyx"
    Laurent de Barry, Co-founder & Chief Sales Officer, Enyx
  "Low-Latency, High Performance Infrastructure for Global Financial Markets"
    Ian Cunningham, Senior Vice President, Europe, Zayo Group
  "FEC off"
    David Riddoch, Chief Architect, Solarflare Communications
  "Distributed Measurement of Packet Latency Variation for Market Data Streams"
    Ron Nevo, VP Systems Engineering, cPacket


Rethinking networks in financeFast Data   

In 2013, Dave Snowdon helped launch Metamako to deliver network hardware solutions to the ultra-low latency market. Five years and many plane trips from Sydney later, Dave and the Metamako team became part of Arista. In this talk, Dave will catch his breath and consider the big picture of networks in financial services enterprises. How do recent business and regulatory trends affect customer requirements around networking? What potential synergies exist between front-office and back-office networks? Dave will offer some opinions on these questions and take more from you.

Innovation RoundupFast Data   Fast Compute   
  "Securing your critical infrastructure with Orolia's Resilient PNT solutions"
    Jeremy Thomas, TSN Market Manager, Orolia
  "Audits v Security – a loser’s game?"
    Steve Newcombe, Account Manager UK, Chronos
  "Integrating high-accuracy synchronization into your financial operations"
    Cesar Prados, Managing Director, Seven Solutions
  "A Machine Learning-Based Approach to Clock Synchronization at Scale"
   Balaji Prabhakar, Co-Founder, Tick Tock Networks


White Rabbit and BeyondFast Data   

Deutsche Boerse have gained valuable experience deploying White Rabbit (roughly PTP with synchronous Ethernet) for time synchronization in their co-location network capture infrastructure. They have also built a data service which makes White Rabbit synchronized timestamps available to exchange members, and they are currently running a pilot project in which members can connect to Deutsche Boerse's White Rabbit master. In this talk, Andreas will discuss lessons learned through these projects. He will also discuss potential plans for further enhancements to the exchange's time synchronization architecture, from improving PTP to using something new and radically different. Come ready to ask questions and toss around ideas.

Plain old PTP: Better than you think?Fast Data   

It is not uncommon to hear that 1PPS doesn't scale well and PTP doesn't give sufficient accuracy for low-latency trading. The implication is that organizations who need scalable distribution of ultra-accurate time need to look beyond these two protocols. Matt has a different view. In particular, he contends that much of criticism lumped on PTP has to do with poor network implementations rather than anything fundamental to do with the protocol. Matt will argue that a well-designed PTP network can deliver time that is accurate to a nanosecond, uses familiar networking components and off-the-shelf (open source) software, and scales well too. In this talk, he will provide both theoretical and empirical evidence to back up this claim. Come to see if Matt can change your views about plain old PTP.